Download The Reaction Mechanism of Silicon

10154
J. Am. Chem. Soc. 1998, 120, 10154-10165
Reaction Mechanism of Silicon-Hydrogen Bond Activation Studied
Using Femtosecond to Nanosecond IR Spectroscopy and Ab Initio
Methods
H. Yang,† M. C. Asplund,§ K. T. Kotz,† M. J. Wilkens,† H. Frei,‡ and C. B. Harris*,†
Contribution from the Department of Chemistry, UniVersity of California, Berkeley, California 94720,
Chemical Sciences DiVision and Physical Biosciences DiVision, MS CalVin Laboratory, Ernest Orlando
Lawrence Berkeley National Laboratory, Berkeley, California 94720
ReceiVed March 2, 1998
Abstract: The Si-H bond activation reactions by group VIIB, d6 organometallic compounds η5-CpM(CO)3
(M ) Mn, Re; Cp ) C5 H5) were studied in neat triethylsilane under ambient conditions. Utilizing femtosecond
and nanosecond pump-probe spectroscopic methods, the spectral evolution of the CO stretching bands was
monitored from 300 fs to tens of microseconds following UV photolysis. The reactive intermediates observed
on the ultrafast time scale were also studied using ab initio quantum chemical modeling. It was found that
photolysis of the manganese tricarbonyl resulted in dicarbonyls in their singlet or triplet electronic states,
whereas photolysis of the rhenium complex led only to the singlet dicarbonyl. The branching ratio of the two
manganese intermediates was measured and was related to different electronic excited states. For both the
Mn and Re complexes, the reactions were found to be divided into two pathways of distinct time scales by the
initial solvation of the dicarbonyls through the Si-H bond or an ethyl group of the solvent molecule. The
time scale for the Si-H bond-breaking process was, for the first time, experimentally derived to be 4.4 ps,
compared to 230 ns for breaking an alkane C-H bond. Knowledge of the elementary reaction steps including
changes in molecular morphology and electronic multiplicity allowed a comprehensive description of the reaction
mechanisms for these reactions.
I. Introduction
The photochemical oxidative addition of a Si-H bond to
certain transition metal complexes has been the focus of many
research efforts since its initial discovery.1 This type of reaction,
which cleaves, or activates, a Si-H bond, is critical in
hydrosilation.2,3 In addition, it provides a direct comparison to
another important bond-cleavage reaction, C-H bond activation
by similar transition metal complexes.4 As such, knowledge
of the reaction mechanism may facilitate the development of
homogeneous catalysis5 and lead ultimately to a general
understanding of the chemistry of bond-cleavage reactions.
Many of the previous studies have used group VIIB, d6
compounds η5-CpM(CO)3 (M ) Mn, Re; Cp ) C5H5) as model
reagents. They provide an interesting contrast of reactivity in
that the extent of the Si-H bond cleavage varies with the type
of metal; the rhenium complex breaks the Si-H bond completely whereas there remains residual Si‚‚‚H interaction in the
* To whom correspondence should be addressed.
† University of California and Chemical Sciences Division, LBNL.
‡ Physical Biosciences Division, MS Calvin Laboratory, LBNL.
§ Present address: Department of Chemistry, University of Pennsylvania.
(1) Jetz, W.; Graham, W. A. G. Inorg. Chem. 1971, 10, 4.
(2) Wrighton, M. S.; Schroeder, M. A. J. Am. Chem. Soc. 1974, 96, 6235.
(3) Schubert, U. AdV. Organomet. Chem. 1990, 30, 151.
(4) (a) Bromberg, S. E.; Yang, H.; Asplund, M. C.; Lian, T.; McNamara,
B. K.; Kotz, K. T.; Yeston, J. S.; Wilkens, M.; Frei, H.; Bergman, R. G.;
Harris, C. B. Science 1997, 278, 260. (b) Arndtsen, B. A.; Bergman, R. G.;
Mobley, T. A.; Peterson, T. H. Acc. Chem. Res. 1995, 28, 154. (c) Bergman,
R. G. Science 1984, 223, 902.
(5) (a) Masters, C. Homogeneous Transition-metal Catalysis; Chapman
Hall: London, 1981. (b) Moser, W. R.; Slocum, D. W. Homogeneous
Transition Metal Catalyzed Reactions; American Chemical Society: Washington, DC, 1992.
manganese case.6 Mechanistically, Young and Wrighton have
established that, at low temperatures, the primary photochemical
product upon UV irradiation of CpMn(CO)3 is the coordinatively
unsaturated, 16-e- CpMn(CO)2 (Scheme 1a).9 Further studies
have shown that the resulting dicarbonyl is complexed by the
solvent to form CpMn(CO)2 (solvent) and that the apparent ratedetermining step involves a loosely solvated transition-state
complex.9-17 The reaction of the rhenium complex CpRe(CO)3
with silanes (Scheme 1b) has also been found to begin with
loss of a CO ligand, followed by solvent coordination. Despite
the many experimental efforts in deriving the reaction mechanism, elementary reaction steps, including the early-time
dynamics and the time scale of the Si-H bond-breaking step,
(6) For example, the distance between the silicon and the hydrogen atom
in the adduct CpRe(CO)2(H)(SiPh3) is 2.19 Å (Ph ) phenyl, C6H5) (ref 7)
compared to 1.80 Å for the similar Mn compound, MeCpMn(CO)2(H)(SiPh2F) (MeCp ) CH3C5H4) (ref 8).
(7) Smith, R. A.; Bennett, M. J. Acta Crystallogr., Sect. B 1977, 33,
1113.
(8) (a) Schubert, U.; Ackermann, K.; Wörle, B. J. Am. Chem. Soc. 1982,
104, 7378. (b) Schubert, U.; Scholz, G.; Müller, J.; Ackermann, K.; Wörle,
B.; Stansfield, R. F. D. J. Organomet. Chem. 1986, 306, 303.
(9) Young, K. M.; Wrighton, M. S. Organometallics 1989, 8, 1063.
(10) Hart-Davis, A. J.; Graham, W. A. G. J. Am. Chem. Soc. 1971, 93,
4388.
(11) Hill, R. H.; Wrighton, M. S. Organometallics 1987, 6, 632.
(12) Creaven, B. S.; Dixon, A. J.; Kelly, J. M.; Long, C.; Poliakoff, M.
Organometallics 1987, 6, 2600.
(13) Klassen, J. K.; Selke, M.; Sorensen, A. A.; Yang, G. K. J. Am.
Chem. Soc. 1990, 112, 1267.
(14) Burkey, T. J. J. Am. Chem. Soc. 1990, 112, 8329.
(15) Hester, D.; Sun, J.; Harper, A. W.; Yang, G. K. J. Am. Chem. Soc.
1992, 114, 5234.
(16) Hu, S.; Farrell, G. J.; Cook, C.; Johnston, R.; Burkey, T. J.
Organometallics 1994, 13, 4127.
(17) Palmer, B. J.; Hill, R. H. Can. J. Chem. 1996, 74, 1959.
S0002-7863(98)00692-1 CCC: $15.00 © 1998 American Chemical Society
Published on Web 09/22/1998
Scheme 1
have remained unattainable due to the extremely rapid reaction
rate.16 Recently, our group has been able to investigate the
room-temperature reaction dynamics of Si-H bond activation
by CpMn(CO)3 in neat Et3SiH (Et ) C2H5) using femtosecond
infrared spectroscopy (ref 18, hereafter denoted as paper I). It
was found that UV excitation of the parent compound CpMn(CO)3 generated two dicarbonyls, and subsequent solvation
partitioned the reaction into two parallel pathways.
In this article, we report the use of a combination of
theoretical modeling and experimental measurements to continue
pursuing this important reaction (Scheme 1). The elementary
reaction steps from initial photolysis to completion of the
reaction were studied to derive a comprehensive reaction
mechanism. Due to the complexity of the reaction and for
clarity, we summarize below some important findings of this
work (cf. Figures 12 and 13) and leave the technicalities for
later sections. The final results can be summarized as follows:
(1) The early-time reaction dynamics depend critically on the
electronic structure of the organometallic complex. For the
manganese complex, there are two overlapping electronic bands
in the accessible UV region (cf. Figure 10). The higher-energy
band is related to formation of an unsolvated singlet dicarbonyl
D* (an asterisk indicates excited state) and the lower one to
triplet dicarbonyl C. A typical UV excitation of the parent
molecule will populate both states, resulting in C or D* through
parallel reaction channels. The branching ratios A f C f D:A
f D* f D for these two parallel channels measured at 295and 325-nm excitations are approximately 50%:50% and 75%:
25%, respectively. For the rhenium complex, only one electronic band appears in the UV region; photolysis of the rhenium
parent molecule leads exclusively to a singlet rhenium dicarbonyl, G*. (2) For both metal complexes, initial solvation of
the nascent coordinatively unsaturated dicarbonyls (C, D*, and
G*), via either the ethyl moiety or the Si-H bond of the solvent
molecule Et3SiH, partitions the reaction into two pathways.
Solvation of the metal dicarbonyl through the Si-H bond leads
(18) Yang, H.; Kotz, K. T.; Asplund, M. C.; Harris, C. B. J. Am. Chem.
Soc. 1997, 119, 9564.
directly to the final products B or F on the ultrafast time scale.
Solvation through an ethyl group results in reactive intermediate
D or G, which reacts further to form the final products in 177
ns and 6.8 µs, respectively. (3) The relative probabilities of
(solvation through the ethyl moiety):(solvation through the Si-H
bond) for the Mn and Re complexes are 5.3 and 3, respectively.
The differences are attributed to steric effects. The larger
diameter of the rhenium atom allows equivalent sampling of
the chemically different sites in a solvent molecule (three ethyl
groups and one Si-H bond). (4) The apparent rate-determining
step for both reactions is the dissociative rearrangement from
the ethyl solvates D or G to the final products. The corresponding free-energy barriers for the manganese and rhenium
complexes, determined from simple transition-state theory, are
8.25 ( 0.03 and 10.41 ( 0.02 kcal/mol, respectively. (5) The
time scale for a Si-H bond-breaking process is, for the first
time, estimated to be 4.4 ( 2.6 ps, indicative of a low-barrier,
if any, reaction.
The paper is organized as follows: The experimental methods
and calculation procedures are described in section II. The
results and detailed data analysis are presented in section III,
in the order of manganese/experimental, rhenium/experimental,
and theoretical results. Implications of the results are discussed
in section IV. In particular, for the manganese complex, we
discuss the effects of sampling different regions of electronically
excited potential energy surface, the nature of the observed transient intermediates, and their interaction with solvent. Comprehensive reaction mechanisms are described in the end of
section IV-A for Mn and in section IV-B for Re. Finally, current
results are compared to the C-H bond activation in section V.
II. Methods
Sample Handling. The compounds η5-CpMn(CO)3 (99%, A) and
η5-CpRe(CO)3 (99%, E) were purchased from Strem, Inc. and used
without further purification. To ensure that moisture and oxygen did
not interfere with the experiment, the triethylsilane acquired from Gelest,
Inc. was dried in an alumina column and distilled under nitrogen. The
sample was prepared under nitrogen atmosphere in an airtight,
demountable liquid IR flow cell (Harrick Scientific Corp.). The
concentration of the CpMn(CO)3/Et3SiH solution was approximately
25 mM. The absorbance of a 400-µm-thick sample of this concentration
at 325 nm was about 1 OD. The concentration of the CpRe(CO)3/Et3SiH solution was approximately 9 mM, to give ∼0.6 OD absorbance
at 295 nm with a 630-µm-thick cell. During the course of experiment,
the static IR spectra were checked regularly to monitor sample
degradation.
Femtosecond Infrared Spectroscopy. Details of the femtosecond
IR (fs-IR) spectrometer setup have been published elsewhere.19 In brief,
the output of a Ti:sapphire oscillator was amplified in two prism-bored
dye-cell amplifiers,20 which were pumped by the output of a frequencydoubled 30-Hz Nd:YAG laser. The amplified light, centered at ∼810
nm, was then split into three beams. One beam was further amplified
to give 70-fs, 20-µJ pulses. The other two beams were focused into
two sapphire windows to generate a white light continuum. Desired
wavelengths of the white light were selected by two band-pass (BP)
filters whose full widths at half-maximum (fwhm) were 10 nm and
further amplified by three-stage dye amplifiers. In this study, the
excitation pulses of 295 and 325 nm were generated by respectively
frequency doubling 590- and 650-nm light. The resulting UV photons
(with energy of ∼6 µJ/pulse) were focused into a disk of ∼200 µm
diameter at the sample to initiate chemical reactions. The required
time delay between a pump pulse and a probe pulse was achieved by
guiding the 590-nm beam (or the 650-nm beam) through a variable
delay line before it was frequency doubled.
(19) Lian, T.; Bromberg, S. E.; Asplund, M. C.; Yang, H.; Harris, C. B.
J. Phys. Chem. 1996, 100, 11994.
(20) Bethune, D. S. Appl. Opt. 1981, 20, 1987.
The broad-band, ultrafast probe pulses centered at ∼5 µm were
generated by mixing a 690-nm with an 810-nm beam. The resulting
∼1-µJ IR pulses, having temporal fwhm of about 70 fs and spectral
bandwidth of about 200 cm-1, were split into signal and reference
beams. Both the signal beam and the reference beam were then focused
into a monochromator and received by a pair of HgCdTe detectors.
The time-resolved IR spectra were collected by scanning the monochromator while fixing the time delay between the pump and the probe
pulses. The kinetic traces were acquired by setting the monochromator
at the desired wavelengths while varying the delay between pump and
probe. The typical spectral and temporal resolutions for this setup are
4 cm-1
Reaction Mechanism of Si-H Bond ActiVation
J. Am. Chem. Soc., Vol. 120, No. 39, 1998 10157
Figure 3. Nanosecond kinetics (O) of CpMn(CO)3 in neat triethylsilane
after 295-nm UV photolysis at (B) 1979 cm-1, the CO stretch of the
silyl adduct CpMn(CO)2(H)(SiEt3) and (D) 1892 cm-1. The finite
intensity of the 1892-cm-1 kinetic trace at longer delay time is probably
due to CpMn(13CO)(12CO)(H)(SiEt3) which has a CO absorption at
∼1887 cm-1.37 The time constants for the exponential fits (dashed
lines) are shown in the graph.
The kinetic trace of D at 1892 cm-1 exhibits a fast rise and
then gradually grows to a constant level. It can be described
by the equation,
Figure 1. Transient difference spectra in the CO stretching region for
CpMn(CO)3 in neat triethylsilane at -10, 10, 33, 66, 200, and 660 ps
following 325-nm UV photolysis. Under the experimental conditions,
the large cross section of the solvent Si-H band (∼2100 cm-1) and
the parent CO bands (1947 and 2028 cm-1) make it difficult to access
some regions of the spectrum (ref 32). The last panel is an FTIR
difference spectrum before and after UV photolysis at 308 nm.
Figure 2. Ultrafast kinetics of CpMn(CO)3 in neat triethylsilane after
325-nm UV photolysis at (B) 1979 cm-1, the CO stretch of the silyl
adduct CpMn(CO)2(H)(SiEt3) (4); (C) 2002 cm-1 (b); and (D) 1892
cm-1 (O). The wavelengths were chosen to minimize overlap with
adjacent peaks. The time constants for the exponential fits (dashed
lines) are shown in the graph.
The peaks at 1892 and 1960 cm-1 are assigned to η5CpMn(CO)2(Et3SiH), denoted D (Scheme 1). Having also been
observed in low-temperature studies,11,16 D is attributed to
solvation of η5-CpMn(CO)2 through the ethyl moiety of Et3SiH. In the 10-ps panel of Figure 1, the two bands at ∼2000
and ∼1883 cm-1, which disappear at longer time scales, are
attributed to a triplet dicarbonyl species, denoted C (see section
IV for assignment). Figure 2 shows the kinetic traces of D, C,
and the final product B following 325-nm excitation. The
kinetics of C recorded at 2000 cm-1 exhibit a rapid rise and an
105 ( 8.0 ps decay.33 The product band at 1979 cm-1 displays
a single-exponential rise of 115 ( 14 ps.
signal(t) ) C1(1 - e-t/τ) + C2
(4)
where τ is a formation time constant, and C1 and C2 are
amplitude constants. This same equation will be used later in
modeling kinetic traces of similar behavior. The fitted numeric
values for τ, C1, and C2 are 107 ( 8.9 ps, 0.73 ( 0.033, and
0.25 ( 0.033, respectively. Clearly, the decay of the intermediate C is correlated to the formation of both the ethyl solvate D
and the final product B. Equation 4 implies two channels for
the formation of the ethyl solvate η5-CpMn(CO)2(Et3SiH), the
branching ratio of which is represented by the relative magnitude
of C1 to C2. With a typical solvation time of a few picoseconds
in mind, the channel that contributes to the fast rise C2 of the
transient of D is attributed to solvation of a singlet η5-CpMn(CO)2, denoted D*. The other channel that contributes to the
107-ps rise is attributed to interconversion from C. Consequently, the branching ratio of the reaction pathway A f C f
D to the pathway A f D* f D in this case is approximately
3:1. The implications of these results will be discussed in
section IV.
The long-time behavior of the reaction was studied using a
nanosecond step-scan FTIR spectrometer. Displayed in Figure
3 are the kinetics of D and that of the product B, recorded at
1892 and 1979 cm-1, respectively. D shows an instrumentlimited rise and then decays with a time constant of 183 ( 8
(32) The extinction coefficients for ν (CO) of CpMn(CO)3 are 12 700
M-1 cm-1 and 5700 M-1 cm-1 for the 1947- and 2028-cm-1 peaks,
respectively (ref 11). To prepare a sample which has an OD ≈ 1 at the
pump wavelength of 325 nm, the IR absorbances of the 1947- and 2028cm-1 bands are calculated to be 12.7 and 5.7, respectively (25 mM, 500µm cell). Similarly, to prepare a sample which has an OD ≈ 0.6 at the
pump wavelength of 295 nm, the IR absorbances of the 1939- and 2030cm-1 bands are calculated to be 9.2 and 3.3 OD, respectively (9 mM, 630µm cell). A thinner cell was used to acquire kinetics for the parent molecule.
Such an experimental difficulty is expected when the sample has a low
absorption cross section in the UV but a high extinction coefficient in the
probe IR region. A sample calculation to estimate the signal strength is
included in ref 36.
(33) Uncertainties represent one standard deviation.
10158 J. Am. Chem. Soc., Vol. 120, No. 39, 1998
Yang et al.
Figure 4. Transient difference spectra in the CO stretching region for
CpRe(CO)3 in neat triethylsilane at -10, 10, 33, 66, 200, and 660 ps
following 295-nm UV photolysis. The arrow at ∼2015 cm-1 indicates
the bleach due to photolysis of the naturally abundant CpRe(12CO)2(13CO).36 Last panel: An FTIR difference spectrum before and after
UV photolysis at 308 nm. Enlarged spectrum (10× and offset by
-0.49) is displayed as a dashed line to show (R) a peak of CpRe(13CO)(12CO)(H)(SiEt3) at ∼1892 cm-1; and (β) bleaches of CpRe(13CO)(12CO)2 at ∼1906 and ∼2019 cm-1.
ns. B exhibits a fast rise and an exponential growth of 177 (
7 ns. The rapid rise of B is from the remnant of the initial
Si-H solvate that forms in the ultrafast time scale. As has been
discussed in paper I, the latter exponential rise of B is due to
the dissociative rearrangement from D.18,34 The nanosecond
kinetics of B can also be modeled with eq 4. In this case, the
coefficients C1 and C2 represent the relative probability of initial
solvation through the ethyl moiety (84%) or the Si-H bond
(16%), respectively. These numbers, however, should be treated
in a qualitative manner. The greater excitation photon flux in
our nanosecond setup compared to the femtosecond one may
cause a transient temperature increase in the sample, which in
turn may result in a time-dependent change in the index of
refraction.35 Such a thermal effect makes it difficult to quantify
the observed kinetics rigorously.
Finally, the free-energy barrier of the dissociative process
from D to B can be derived from the kinetic traces. If one
assumes a simple transition-state theory, the reaction rate can
be expressed as the reciprocal of the observed lifetime τ,
1 kBT -∆Gq/kBT
reaction rate ≈ )
e
τ
h
(5)
where kB is the Boltzmann constant, h Plank’s constant, T room
temperature (298.15 K), and ∆Gq the free energy of activation.
The reaction barrier, estimated from the product formation time,
is ∆Gq ) 8.25 ( 0.03 kcal/mol. This is in excellent agreement
with the literature value, 8.56 ( 0.49 kcal/mol, which is
(34) For a recent discussion on the “chain-walk” mechanism vs the
“dissociative” mechanism, see: Ladogana, S.; Nayak, S. K.; Smit, J. P.;
Dobson, G. R. Inorg. Chem. 1997, 36, 650.
(35) Yuzawa, T.; Kato, C.; George, M. W.; Hamaguchi, H.-O. Appl.
Spectrosc. 1994, 48, 684.
extrapolated to room temperature using the activation parameters
obtained from low-temperature measurements.11
B. Activation of the Silicon-Hydrogen Bond of Et3SiH
by CpRe(CO)3. Displayed in Figure 4 are the transient fs-IR
spectra of CpRe(CO)3 in neat Et3SiH following excitation at
295 nm. The bottom panel is an FTIR difference spectrum taken
after prolonged exposure to 308-nm light. The presence of
naturally abundant 13CO stretches is evidenced by the enlarged
FTIR spectrum.36 The small peak at 1892 cm-1 is assigned to
η5-CpRe(13CO)(12CO)(H)(SiEt3), denoted R in the enlarged
spectrum. The shallow bleaches labeled as β are due to
depletion of naturally abundant η5-CpRe(12CO)2(13CO).37 At
the 10-ps panel, the pair of bands at 1887 and 1948 cm-1 are
assigned to η5-CpRe(CO)2(Et3SiH) [G], resulting from solvation
of the η5-CpRe(CO)2 [G*] through the ethyl moiety of the
solvent.11 The final product F appears as two bands at 1918
and 1990 cm-1, but its early-time spectra are complicated by
absorptions from vibrationally excited parent molecules E (the
10- and 33-ps panels).
As shown in Figure 5a, the kinetics of the parent molecule
E taken at 2030 cm-1 display an instrument-limited depletion
which recovers with a time constant of 39.3 ( 1.6 ps to 28%
of its initial intensity, consistent with the literature quantum
yields of 0.30 at 313-nm excitation.38 The recovery of the parent
bleach is attributed to reestablishing the thermal distribution of
vibrational states in the electronic-ground manifold.39-41,43
(36) In the following, we will use the Re system as an example to estimate
the signal size of F in the ultrafast spectra. The same calculation is
applicable to Mn spectra. We start by calculating the OD change in one of
the parent bands, e.g., the 2030-cm-1 band. The pump beam consists of
295-nm pulses of energy ∼6 µJ, 75% of which are absorbed by an OD ≈
0.6 sample. Change in the parent absorbance is calculated by ∆OD(E) )
l∆cparent, where l is the sample cell thickness and ∆cparent is the concentration
change of the parent species in the cylindric volume defined by the pump
beam. Taking into account the cell thickness (650 µm), the beam size at
the sample (∼200 µm), and quantum yield (∼0.3), each pump pulse induces
OD change in the parent bleach of ∼0.06. The signal size of F on the
ultrafast time scale can be expressed as
∆OD(F) ) ∆OD(E) × X(F) ×
(F)
(E)
where X(F) ) 0.248 ( 0.079 is the branching percentage for F measured
from the nanosecond kinetics, and the extinction coefficient ratio (F)/(E) ≈ 2/5 measured from the static difference FTIR spectrum on the bottom
panel of Figure 1. ∆OD(F) is thus calculated to be ∼0.006, which is
consistent with what we have measured in the present work. The above
calculation also serves as a cross-examination for the measured 25%
branching ratio in the initial solvation through the Si-H bond of Et3SiH.
The recorded 13C bleach signal in the ultrafast spectra of the Re complex
may appear somewhat larger than one would have expected from a simple
estimate. In the following, we show that the Re spectra displayed reasonable
13C bleach at 2015 cm-1. Take the 200 ps panel of Figure 4 as an example.
The numerical values for the 1990-cm-1 band of F and the 13C bleach at
2015 cm-1 are about 0.0053 ( 0.0005 OD and -0.0036 ( 0.001 OD,
respectively. Assuming 3% natural abundance of 13C, and taking into
account the difference in extinction coefficients (F)/(13C) ≈ 2/5 and
branching ratios 0.248 ( 0.079, one finds that the estimated signal size of
the 13C bleach is ∼ -0.0017 ( 0.0005 OD, within a 95% confidence interval
compared to recorded signal size (-0.0036 ( 0.001 OD). Similarly, the
corresponding 13C bleach of Mn complex has a signal size of -0.0027 (
0.0006 OD (taken from the 660-ps panel of Figure 1), which is also
reasonable within experimental error.
(37) CO frequencies of CpRe(13CO)x(12CO)3-x in room-temperature
alkane solutions were estimated from an effective force field calculation
following the methods in: (a) Haas, H.; Sheline, R. K. J. Chem. Phys.
1967, 47, 2996. (b) Wilson, E. B., Jr.; Decius, J. C.; Cross, P. C. Molecular
Vibrations; Dover: New York, 1980. The calculated frequencies (in cm -1)
are CpRe(12CO)3 [1939, 2030]; CpRe(12CO)2(13CO) [1906, 1939, 2019];
CpRe(12CO)(13CO)2 [1895, 1919, 2005]; and CpRe(13CO)3 [1895, 1985].
Similarly, peaks originating from CpM(13CO)(12CO)(H)(SiEt3) are also
estimated to be at 1892 and 1974 cm-1 for the rhenium complex and at
1887 and 1964 cm-1 for the manganese complex.
(38) Giodano, P. J.; Wrighton, M. S. Inorg. Chem. 1976, 16, 160.
(39) King, J. C.; Zhang, J. Z.; Schwartz, B. J.; Harris, C. B. J. Chem.
Phys. 1993, 99, 7595 and references therein.
Processes that contribute to this nonequilibrated vibrational
distribution include depletion of the V ) 0 state by the pump
photon and population replenished by nondissociative processes.
As a consequence of populating the higher vibrational states,
transitions such as V ) 1 f 2 and V ) 2 f 3 would become
detectable. The broad band at ∼1921 cm-1 is assigned to the
V ) 1 f 2 absorption of the 1939-cm-1 CO stretching band of
E, which evolves to the blue end of the spectrum and decays
completely at later time (10-ps panel, Figure 4). The observed
∼18-cm-1 anharmonicity is consistent with the reported anharmonicity of other metal carbonyls.40-43 Similarly, the broad
bands at ∼2004 and ∼1990 cm-1 in the 10-ps panel of Figure
4 can be assigned to the V ) 1 f 2 and V ) 2 f 3 absorption
of naturally abundant CpRe(13CO)(CO)2, respectively. The
parent molecules, containing one 13CO ligand, manifest themselves as a depletion at 2015 cm-1, as indicated by the arrow
in the 660-ps panel of Figure 4 [see ref 37 for estimation of
CO stretching frequencies of CpRe(13CO)x(12CO)3-x]. Unfortunately, this 2015-cm-1 band is superimposed on one of the
parent CpRe(12CO)3 V ) 1 f 2 transition. Therefore, instead
of showing an instantaneous depletion followed by a rapid
recovery, it bleaches gradually to a constant level (Figure 6c).
The vibrationally hot bands described above unavoidably
impede an accurate description of bands that are important in
determining the reaction dynamics. The 1990-cm-1 band of
the final product F overlaps with the V ) 2 hot band (at ∼1900
cm-1) of the CpRe(13CO)(CO)2 bleach at 2015 cm-1. As a
result, the kinetic trace of F measured at 1990 cm-1 exhibits a
sharp rise due to the hot band and then decays to a constant
level (Figure 6). To derive the underlying product dynamics,
the kinetic traces measured at 1990, 2004, and 2015 cm-1 were
fit to a model which takes into account the bleach recovery,
product formation, and vibrational relaxation (see Appendix).
The extracted formation time for the product is 4.4 ( 2.6 ps.
10160 J. Am. Chem. Soc., Vol. 120, No. 39, 1998
Figure 7. Nanosecond kinetics (O) of CpRe(CO)3 in neat triethylsilane
after 295-nm UV photolysis at (F) 1990 cm-1, the CO stretch of the
silyl adduct CpRe(CO)2(H)(SiEt3), and (G) 1950 cm-1, the CO stretch
of the alkyl solvate CpRe(CO)2(Et3SiH). The time constants for the
exponential fits (dashed lines) are shown in the graph. Inset: A short
scan at 1990 cm-1 showing the instrument-limited rise due to formation
of product F on the ultrafast time scale.
0.2 µs exponential decay to a constant level. This finite
absorbance is attributed to the CO stretch of CpRe(13CO)(12CO)(H)(SiEt3), estimated to be at 1892 cm-1, the kinetics of which
should be similar to those of the product.37 The alkyl solvate
G, therefore, displays a slightly longer decay time than the rise
time of F. Similar to the treatment performed in the CpMn(CO)3/Et3SiH system, the relative probability of the initial
solvation through the ethyl group and the Si-H bond is
estimated to be 75% and 25%, respectively. Assuming that
transition-state theory is applicable, the measured rise time of
the product gives an estimate of the free energy of the reaction
barrier to be ∆Gq ) 10.41 ( 0.02 kcal/mol (eq 5). Thus, the
apparent rate-determining step is shown to be the dissociative
rearrangement of the initial alkyl solvate, similar to the
manganese reaction.
C. Ab Initio Calculation Results. Geometry Optimization. Figure 8 shows the geometries of the parent molecules
η5-CpM(CO)3,44 the dicarbonylmetal complexes solvated by an
ethane molecule η5-CpM(CO)2 (ethane), and the final products
η5-CpM(CO)2(H)(SiH2CH3), all optimized at the MP2/lanl2dz
level of theory. As displayed in Table 1, the metal-Cp and
metal-CO distances for the parent molecules calculated at the
MP2 level agree reasonably well with the experimental values,
except for the slightly overestimated C-O bond lengths.
Qualitatively, the metal-ligand distances should reflect the
relative sizes of the manganese and the rhenium atoms. As
displayed in Table 1, the rhenium complexes exhibit consistently
a longer metal-ligand distance by ∼0.17 Å over the manganese
complexes, indicating a larger diameter for a complexed Re
atom over a Mn atom.
For the ethane dicarbonyl complexes in Figure 8, the ethane
molecule interacts with the metal center via one of its C-H
bonds through an η2 interaction. The bond lengths for the
coupling C-H bond are 1.14 and 1.17 Å for the Mn and Re
compounds, respectively. These calculated C-H bond lengths,
compared to 1.10 Å for a typical C-H bond in a saturated
alkane from the same level of calculation, show a significantly
(44) The MacMolPlt package used for this graphical illustration
was obtained from http://www.msg.ameslab.gov/GAMESS/Graphics/
MacMolPlt.shtml.
Yang et al.
Figure 8. Geometries of the parent molecules, the ethane solvate, and
the final product optimized at the MP2/lanl2dz level of theory. The
bond lengths are in angstroms and angles in degrees. Parameters for
the rhenium complexes are in parentheses.
extended C-H bond (+0.07 Å) when complexed to the Re
center but only slight (+0.04 Å) extension when coupled to
the Mn complex. The more extended C-H bond in the Re
case suggests that the Re atom interacts more strongly with a
C-H bond, hence forming a more stable alkyl complex.
Finally, the relative extent of the Si-H bond activation by the
Mn and Re compounds is clearly demonstrated by the Si-H
bond distances in the final products. The calculated Si-H bond
lengths for the Mn and Re complexes are respectively 1.90 and
2.35 Å, compared to a typical silane Si-H bond length of 1.49
Å. Apparently, the degree of bond-breakage can be inferred
from the product Si-H bond lengths. In the hydridosilyl
manganese product, the Si-H bond is 0.41 Å longer than a
Si-H bond in an isolated silane. The hydridosilyl rhenium
product, on the other hand, exhibits a much longer distance (0.86
Å longer than an isolated Si-H bond) between the silicon and
the hydrogen atoms, marking a completely broken Si-H bond.
Shown in Figure 9 are the geometries for the 16-e- species
η5-CpM(CO)2 in their singlet and triplet states. Despite the
similarity of the structures, one notices a difference in the OCM-CO angle. Due to the larger radius of the Re atom and,
hence, reduced ligand-ligand repulsive interaction, the angle
is 93.47° for the Re complex, compared to 95.11° for the Mn
one or, when projected onto the Cp plane, 121° for the Re and
128° for the Mn species.
For both metal complexes, the geometries for the singlet
species are quite different from the triplet ones; most notable
are the positions of the M-CO bond relative to the molecular
C2V plane. For example, the angle between the M-CO bond
and the molecular C2V plane for the singlet state η5-CpMn(CO)2
is 64°, but it is 87° for the triplet state. In addition, the M-Cp
and M-CO distances of the triplet species are longer than those
of the singlet ones. Also, the CO bond lengths are shorter in
the triplet states compared to those of the singlet states. Such
a decrease in the metal-CO interaction in the triplet state has
also been observed in Siegbahn’s calculations of CpM(CO) (M
) Co, Rh, and Ir).45 Also shown in Table 1 are the DFT results
for the 16-e- species which are in general agreement with those
(45) Siegbahn, P. E. M. J. Am. Chem. Soc. 1996, 118, 1487.
Reaction Mechanism of Si-H Bond ActiVation
J. Am. Chem. Soc., Vol. 120, No. 39, 1998 10161
Table 1. Critical Bond Distances (Å) of the Important Structures for the Si-H Bond Activation Reactions Optimized at the MP2/lanl2dz
Level of Theorya
compound
Cp-M
M-CO
C-O
η5-CpMn(CO)3
singlet η5-CpMn(CO)2
triplet η5-CpMn(CO)2
η5-CpMn(CO)2(H3CCH3)
η5-CpMn(CO)2(H)(SiH2CH3)
η5-CpRe(CO)3
singlet η5-CpRe(CO)2
triplet η5-CpRe(CO)2
η5-CpRe(CO)2(H3CCH3)
η5-CpRe(CO)2(H)(SiH2CH3)
1.77 (1.77)b
1.80 [1.86]
1.84 [1.94]
1.79
1.75
2.02 (1.96)c
1.94 [1.97]
2.03 [2.04]
1.96
2.00
1.74 (1.78)b
1.78 [1.79]
1.82 [1.83]
1.77
1.73
1.96 (1.90)c
1.95 [1.89]
1.98 [1.91]
1.95
1.97
1.22 (1.16)b
1.21 [1.19]
1.19 [1.18]
1.22
1.22
1.21 (1.17)c
1.22 [1.19]
1.19 [1.19]
1.21
1.21
M-H
M-Si(C)
Si-H (C-H)
1.77
1.51
2.73
2.44
1.14
1.90
1.89
1.69
2.84
2.57
1.17
2.35
a Numbers in parentheses are experimental values, and those in square brackets are from DFT calculations. b Fitzpatrick, P. J.; Le Page, Y.;
Sedman, J.; Butler, I. S. Inorg. Chem. 1981, 20, 2852. c Fitzpatrick, P. J.; Le Page, Y.; Sedman, J.; Butler, I. S., Acta Crystallogr., Sect. B 1981,
B37, 1052.
Figure 9. Side view and top view for the MP2/lanl2dz optimized
geometries of the 16-e- species. The bond lengths are in angstroms
and angles in degrees. Parameters for the rhenium complexes are in
parentheses.
from the MP2 calculations. The DFT-calculated angle between
the M-CO bond and the molecular C2V plane for the singletstate η5-CpMn(CO)2 is 63°, and it is 84° for the triplet state.
Analogous parameters for the singlet and triplet rhenium
complexes are 58° and 88°, respectively. The C-O bond
lengths predicted by DFT methods, however, are consistently
shorter than those predicted by MP2 methods. The Cp-Mn
distance from DFT calculations is significantly longer than that
from UMP2 methods (by ∼0.1 Å), perhaps due to the
inadequacy of the UMP2 wave functions for describing highspin species. These geometric parameters are summarized in
Table 1. Several attempts to optimize η3-CpMn(CO)2, η3CpMn(CO)2(CH4), or η3-CpMn(CO)2(SiH4) were unsuccessful,
however, as an η5-CpMn(CO)2 configuration was invariably
obtained, regardless of the initial guess.
Energy Calculations. The calculated metal-ligand interaction energies are listed in Table 2. After BSSE and ZPE
corrections, the binding energies between the Mn complex and
the CO ligand or the Si-H bond appear to be in good agreement
with the experimental values. Compared to the Mn complex,
the Re complex shows a binding energy to the CO ligand which
is stronger by 5.81 kcal/mol. As a result, there is comparatively
less excess energy available to the Re metal fragment after
photolysis. Also, the hydridosilyl rhenium product is thermodynamically more stable than the manganese product by a
marked 11.98 kcal/mol. In the case of an alkane ligand, it has
been found experimentally46 and theoretically47 that the interaction energy between a transition metal and a series of chain
alkanes may increase with alkane chain length. Consequently,
in the present calculation, which uses ethane as the complexed
solvent to model the solvation of η5-CpM(CO)2 through the ethyl
group of Et3SiH, the complexation energies obtained are likely
to be underestimated. Nonetheless, the calculated binding
energies of the ethane with Mn and Re complexes allow a
qualitative understanding of the relative stability of η5-CpMn(CO)2 (alkane) and η5-CpRe(CO)2 (alkane). Our results show
that an ethane molecule binds more strongly to the Re complex
than to the Mn complex by about 3.5 kcal/mol. Similar stability
for the η5-CpRe(CO)2L (L ) n-heptane, Xe, and Kr) has also
been reported in a recent paper by Sun et al.48 Consequently,
if the free-energy barrier of a dissociative rearrangement is
dominated by the enthalpy of complexation, it would take longer
for an η5-CpRe(CO)2(Et3SiH) to rearrange dissociatively to
allow interaction of a Si-H bond with the metal center, in
comparison with the manganese system. Our experimental
results show that it takes 6.8 µs for the rhenium compound to
undergo such a rearrangement, whereas only 177 ns is required
for the analogous process involving the manganese compound
or, equivalently, a 2.16 kcal/mol difference in their free-energy
barrier.
We next turn to the relative energies of the singlet and the
triplet η5-CpMn(CO)2. The DFT results show a lower energy
for the triplet η5-CpMn(CO)2 relative to the singlet (∆E ) 8.09
kcal/mol), in contrast to the MP2 results, which predict a lowerenergy singlet state by 10.93 kcal/mol. These conflicting results
are probably due to the near-degeneracy of the singlet and the
triplet states. Similar inconsistent results from different calculation methods have also been reported in the literature.45 For
CpCo(CO), Siegbahn found that the MP2 method gave a singlet
ground state, whereas other methods, including DFT/B3LYP
and CASPT2, consistently yield a triplet ground state. To
minimize the near-degeneracy effect, CAS MCSCF-PT2 calculations were performed at the DFT geometries; the results
predict a ground triplet state for η5-CpMn(CO)2, 9.86 kcal/mol
lower than the singlet state. Therefore, our calculations are in
favor of a triplet ground state. For the rhenium complexes,
analogous calculations consistently predict a singlet ground state
(46) (a) Ishikawa, Y.; Brown, C. E.; Hackett, P. A.; Rayner, D. M. Chem.
Phys. Lett. 1988, 150, 506. (b) Brown, C. E.; Ishikawa, Y.; Hackett, P. A.;
Rayner, D. M. J. Am. Chem. Soc. 1990, 112, 2530.
(47) Zaric, S.; Hall, M. B. J. Phys. Chem. A 1997, 101, 4646.
(48) Sun, X.-Z.; Grills, D. C.; Nikiforov, S. M.; Poliakoff, M.; George,
M. W. J. Am. Chem. Soc. 1997, 119, 7521.
10162 J. Am. Chem. Soc., Vol. 120, No. 39, 1998
Yang et al.
Calculated MP2/lanl2dz Binding Energies (kcal/mol) for Singlet η5-CpM(CO)2L, M ) Mn, Rea
Table 2.
Mn
Re
L
ZPE
∆E
∆E(BSSE)
ZPE
∆E
∆E(BSSE)
CO
Et
SiH3CH3
-2.26b
0.21d
111.93d
95.55
16.21
53.83
51.76 (55)c
2.91 (10)e
22.92 (24.4)e
-2.75b
0.21d
109.62d
97.87
20.56
71.88
57.57
6.41
34.90
a Experimental values are in parentheses. b Calculated using HF frequencies. c Angelici, R. J.; Loewen, W. Inorg. Chem. 1967, 6, 682. d Calculated
using MP2 frequencies. e Burkey, T. J. J. Am. Chem. Soc. 1990, 112, 8329.
Table 3. Comparison of the Energy Difference, ∆E ) E(singlet)
- E(triplet) (kcal/mol), between the Singlet and Triplet
η5-CpM(CO)2 Calculated Using Different Methods
MP2a
DFTb
CASSCF-PT2b
Mn
Re
-10.93
8.09
9.86
-5.63
-5.62
-7.88
a
Calculated at the MP2/lanl2dz optimized geometry. b Calculated
at the B3LYP/lanl2dz optimized geometry.
Figure 10. UV/vis spectra of the CpMn(CO)3 and CpRe(CO)3 in neat
triethylsilane taken under experimental condition. Excitation wavelengths are indicated by arrows.
for η5-CpRe(CO)2. These results are summarized in Table 3.
Finally, numerical CO stretching frequencies at DFT level were
calculated for the singlet η5-CpMn(CO)2 (methane) (1889 and
1946 cm-1) and for the triplet η5-CpMn(CO)2 (1870 and 1967
cm-1). These results will be used in the assignment of C in
the next section.49
IV. Discussion
A. Activation of the Silicon-Hydrogen Bond of Et3SiH
by CpMn(CO)3. Effects of Excitation Wavelength. In the
case of 295-nm excitation (paper I), the parent molecule η5CpMn(CO)3 loses one CO ligand to produce two dicarbonyls
(designated as D* and C in Figure 12); the former is quickly
solvated to give D. The intermediate C then converts to form
D on a time scale of 90 ps, or to B in ∼71 ps. The branching
ratio in generating these two intermediates at this excitation
wavelength, calculated in paper I from the kinetic trace of D,
is approximately 1:1. Since this pump energy resides in the
trough of two electronic states (see Figure 10), it is conceivable
that D* and C may be related to the two electronically excited
states. To examine how the excited states affect the reaction
mechanism, we changed the excitation wavelength to the lowerenergy band at 325 nm in the current study. Compared to
excitation at 295 nm, the 325-nm excitation light injects less
energy to the chemical system and samples different regions of
the excited potential energy surface (PES). As a result, it may
(49) The frequencies for a triplet dicarbonyl were determined on the η5CpMn(CO)2 instead of η5-CpMn(CO)2 (alkane) because it is only weakly
solvated by alkanes, to be discussed in the next section.
Figure 11. Ultrafast kinetic traces of D (2), C (0) following 295-nm
excitation, and D (O) following 325-nm excitation. Also shown are
the functions Y295 and Y325 that are used to model the two normalized
traces of D. Corresponding fits are shown as dashed lines. Clearly,
the relative intensity of C to D is lower when excited at 295 nm
compared to that at 325-nm excitation in Figure 2. This wavelength
dependence can be quantified by comparing the fitting coefficients of
Y295 and Y325. See text for details.
lead to dissociation channels that have different product energy
distribution. Indeed, the kinetics of C and D following 325nm excitation show little, if any, excess vibrational energy. This
is to be compared with results from 295-nm excitation, which
indicate that ∼23 ps is required to dissipate excess vibrational
energy. For clarity, the kinetic traces of D recorded at the two
pump energies are reproduced in Figure 11. Accessing a
different part of the excited PES also changes the branching
ratio from A f D* f D:A f C f D ≈ 1:1 (295-nm excitation)
to 1:3 (325-nm excitation). This may imply that the lowerenergy band at ∼330 nm correlates with the dissociation channel
that leads to C and the higher-energy band to D.
Nature of the Intermediate C. In paper I, two possible
candidates for the observed transient intermediate C were
proposed: a ring-slipped η3-CpMn(CO)2 or η5-CpMn(CO)2 in
another electronic state, presumably a triplet state. The η3CpMn(CO)2 species,50 however, has only 14 valence electrons
for the Mnsa first-row transition metalsseverely violating the
18-e- rule. Our theoretical modelings also support the energetic
instability of η3 complexes. C is therefore unlikely to be a ringslipped 14-e- species. Another possible assignment for C is
that it is a dicarbonyl in the triplet ground state. There have
been precedents for 16-e- organometallics having a triplet
ground state,51-54 the geometry of which may differ from that
(50) (a) O’Connor, J. M.; Casey, C. P. Chem. ReV. 1987, 87, 307. (b)
Lees, A. J.; Purwoko, A. A. Coord. Chem. ReV. 1994, 132, 155. (c) Basolo,
F. New J. Chem. 1994, 18, 19.
(51) Bengali, A. A.; Bergman, R. G.; Moore, C. B. J. Am. Chem. Soc.
1995, 117, 3879.
Reaction Mechanism of Si-H Bond ActiVation
of the singlet state to allow a different ligand vibrational
frequency.55 Our calculations suggest a triplet ground state for
η5-CpMn(CO)2. We next compare the theoretical CO stretching
frequencies to the experimental results. Recent developments
in ab initio methods have shown promising results in identifying
organometallic species through frequency calculations.56 The
DFT calculations for the triplet dicarbonyl show a +21-cm-1
shift of the higher-energy CO band and a -19-cm-1 shift of
the lower-energy band of the solvated singlet dicarbonyl,
compatible with experimental observation of frequency shifts
of C relative to D. Therefore, it seems reasonable to attribute
C to a triplet η5-CpMn(CO)2.
Solvation of the Triplet and Singlet Manganese Dicarbonyls. In the context of C being a triplet-state η5-CpMn(CO)2,
we next discuss the interaction of C and the solvent. Unlike
the singlet 16-e- species, which can be solvated readily by
alkanes and rare-gas atoms (Kr, Xe), it has been suggested
theoretically that transition metals in their triplet state interact
only weakly with these solvents.57 This conclusion has been
used to explain the high reactivity of triplet η5-CpCo(CO)
toward a free CO in liquefied rare-gas solution or in alkane
solution.51 In that case, the incoming CO ligand does not require
much activation energy to displace an alkane or rare-gas
molecule which couples weakly to the metal center of η5-CpCo(CO). Triplet-to-singlet intersystem crossing in chemical
systems such as triplet η5-CpCo(CO) or triplet η5-CpMn(CO)2
is further complicated by the fact that the complex of interest
is always under the influence of solvent molecules in a liquid
environment. It was found by Dougherty and Heilweil that
triplet η5-CpCo(CO) reacts very quickly (<vibrational cooling
time) with the strongly binding ligand 1-hexene to form
presumably a singlet π-complex, in which a 1-hexene molecule
complexes to the Co metal through its CdC double bond.43
This is to be compared with the lifetime (∼milliseconds) of
triplet η5-CpCo(CO) in weakly binding solvents such as Kr and
Xe.51 Hence, the surrounding solvent molecules, especially
those interacting strongly with the metal center, need to be
considered in describing the reaction coordinate for spincrossover and solvent exchange. In other words, the interconversion from triplet C to singlet D should probably be considered
as a concerted spin-crossover/solvation process.
With this picture in mind, the solvation of the initially
photogenerated triplet and singlet dicarbonyls can be described
as follows. Probably due to steric hindrance and statistical favor
of the ethyl groups over the Si-H bond, the nascent singlet
η5-CpMn(CO)2 is preferentially solvated via the ethyl moiety
of Et3SiH to form D. The solvation of the triplet dicarbonyl,
however, may be quite different. Assuming that a triplet η5CpMn(CO)2 interacts initially with the ethyl moiety of an Et3SiH molecule, the interaction between the alkyl group and a
triplet Mn is so weak that the surrounding solvent molecules
have sufficient thermal energy to easily reorient or displace the
(52) Abugideiri, F.; Keogh, D. W.; Poli, R. J. Chem. Soc., Chem.
Commun. 1994 2317.
(53) Detrich, J. L.; Reinaud, O. M.; Rheingold, A. L.; Theopold, K. H.
J. Am. Chem. Soc. 1995, 117, 11745.
(54) Poliakoff, M.; Weitz, E. Acc. Chem. Res. 1987, 20 408.
(55) (a) Rawlins, K.A.; Lees, A. J. Inorg. Chem. 1989, 28, 2154. (b)
Servaas, P. C.; van Dijk, H. K.; Snoeck, T. L.; Stuhkens, D. J.; Oskam, A.
Inorg. Chem. 1985, 24, 4494.
(56) (a) Jonas, V.; Thiel, W. J. Chem. Phys. 1996, 105, 3636. (b) Zaric,
S.; Couty, M.; Hall, M. B. J. Am. Chem. Soc. 1997, 119, 2885. (c) Zaric,
S.; Hall, M. B. J. Phys. Chem. 1998, 102, 1963.
(57) (a) Blomberg, M. R. A.; Siegbahn, P. E. M.; Svensson, M. J. Am.
Chem. Soc. 1992, 114, 6095. (b) Siegbahn, P. E. M.; Svensson, M. J. Am.
Chem. Soc. 1994, 116, 10124. (c) Carroll, J. J.; Weisshaar, J. C.; Haug, K.
L.; Blomberg, M. R. A.; Siegbahn, P. E. M.; Svensson, M. J. Phys. Chem.
1995, 99, 13955.
J. Am. Chem. Soc., Vol. 120, No. 39, 1998 10163
Figure 12. Proposed reaction mechanism for the silicon-hydrogen
bond activation by CpMn(CO)3 covering the ultrafast dynamics to
nanosecond kinetics. The nascent singlet η5-CpMn(CO)2, denoted D*,
is also shown for the purpose of discussion. See text for details.
interacting Et3SiH, until the triplet Mn encounters a strongly
coupling Si-H site. This process is expected to be on the order
of a few picoseconds, the time scale for typical solvent motions.
It would thus appear that a triplet transition metal complex is
preferentially solvated by the more strongly coupling sites of a
solvent molecule. Similarly a preferential solvation has also
been observed in Heilweil’s experiments on CpCo(CO) in neat
1-hexene, where one would expect to see solvation of CpCo(CO) through both the alkyl chain and the CdC bond of
1-hexene. The only product observed, however, was the
strongly bound π-complex. Our data show that the product
kinetics exhibit only one exponential rise on the ultrafast time
scale, which is perhaps due to the preferential initial solvation
of the singlet η5-CpMn(CO)2 via the ethyl group and the triplet
η5-CpMn(CO)2 through the stronger-binding Si-H bond of the
solvent molecule.
The Reaction Mechanism. In light of the above discussion,
the photochemical Si-H bond activation reaction of η5-CpMn(CO)3 in neat Et3SiH under the ambient conditions is summarized in Figure 12. We found that there are two excited states
involved in the experimentally accessible region. The higherenergy UV band is correlated with the generation of the
dicarbonyl η5-CpMn(CO)2(D*) in its electronic singlet manifold,
while the lower-energy band is correlated to the triplet η5-CpMn(CO)2 (C). Following the previous discussion on the preferential
solvation of D* through the ethyl moiety and the fact that 84%
of the total product comes from D, the branching ratio for A f
D* to A f C can be calculated to be 42%:58% at 295-nm
excitation. The initial solvation of these two intermediates via
the Si-H bond or the ethyl moiety of the solvent partitions the
reaction into two pathways with a branching ratio close to 1:5.3
(16% through the Si-H bond and 84% through the ethyl group).
Solvation of C via the Si-H bond leads to the final product B
on a time scale of ∼100 ps, possibly through a concerted
coordination/spin-crossover process. Solvation of D* through
the ethyl group of the solvent molecule results in D, whose
population is augmented by interconversion from the triplet C
on a similar time scale of ∼100 ps. Finally, the ethyl solvate
reacts to form the final product B through a dissociative
10164 J. Am. Chem. Soc., Vol. 120, No. 39, 1998
Figure 13. Proposed reaction mechanism for the silicon-hydrogen
bond activation by CpRe(CO)3 covering the ultrafast dynamics to
microsecond kinetics. The nascent singlet and triplet η5-CpRe(CO)2,
denoted H* and G*, are also shown for the purpose of discussion. See
text for details.
rearrangement on a time scale of ∼177 ns, corresponding to a
barrier height of ∆Gq ≈ 8.25 kcal/mol.
B. Activation of the Silicon-Hydrogen Bond of Et3SiH
by CpRe(CO)3. In comparison to the manganese compound,
the photochemical reaction of η5-CpRe(CO)3 (E) with Et3SiH
is much simpler (see Figure 13). Photolysis of E results in a
dicarbonyl species, η5-CpRe(CO)2 (G*), in its singlet electronic
ground state. Subsequent solvation of G* partitions the reaction
into two courses to the formation of the final product. Interaction of G* with the Si-H bond of the solvent molecule leads
directly to the final product η5-CpRe(CO)2(H)(SiEt3) (F), with
a time constant of ∼4.4 ps. Another pathway proceeds through
solvation of G* via the ethyl group of the solvent to form η5CpRe(CO)2(Et3SiH) (G) on a time scale of ∼2.5 ps and further
reacts to form the final product F. The relative probability of
(solvation through the ethyl moiety)/(solvation through the Si-H
bond) for the rhenium complex is found to be approximately 3,
compared to 5.3 for the manganese complex. The difference
may be attributed to steric interaction from the ligands. The
smaller radius of the Mn center may cause increased steric
interaction between the ligands and the solvent molecule’s ethyl
group when solvation proceeds through the Si-H bond. Such
steric hindrance is reduced when solvation occurs through the
ethyl moiety. The larger diameter of the Re center makes it
more accessible to chemically different sites of a solvent molecule, resulting in a branching ratio of approximately 3sthe ratio
of three ethyl groups to one Si-H bond in an Et3SiH molecule.
V. Closing Remarks
Due to the very complicated dynamic processes in a liquid
environment, photochemical reactions such as the bond-activation reactions studied here may span several orders of magnitude
in timesfrom hundreds of femtosecond for dissociating a
chemical bond to the time scale governed by a diffusional
process or an energetic barrier. Combining the use of the
femtosecond and nanosecond UV pump-IR probe spectroscopy,
Yang et al.
the elementary reaction steps including changes in molecular
conformation and electronic multiplicity may be studied by
monitoring the time evolution of the reactive intermediates. The
nature of the intermediates whose lifetimes were too short for
conventional spectroscopic characterization were studied using
first-principle quantum chemical methods. Utilizing these
powerful tools, it is also possible to assess the relative probability
of parallel, dynamically partitioned reaction channels. We have
investigated photochemical Si-H bond activation reaction by
group VIIB, d6 transition metal complexes η5-CpM(CO)3 (M
) Mn, Re). The detailed reaction mechanisms are summarized
in Figures 12 and 13.
Despite the fact that Mn and Re belong to the same group in
the periodic table, we have shown in this paper that the reaction
dynamics of η5-CpMn(CO)3 or η5-CpRe(CO)3 with Et3SiH is
very different on the ultrafast time scale and depends on the
nature of electronic structure of the molecule in question. If
more than one dissociation channel (excited state) is involved
in the initial photolysis, as in the Mn case, which has two
experimentally accessible UV bands, the reaction may dynamically partition into several parallel pathways leading to different
transient intermediates. The nature of these transient species
have been investigated using ab initio calculations. The
calculation results provided strong evidence that photolysis of
the parent η5-CpMn(CO)3 generates both singlet and triplet
CpMn(CO)2 species, which accounts for the experimental
observations consistently. The correlation of these reaction
pathways to the excited states was studied by examination of
their branching ratios, which were determined from ultrafast
kinetic traces of the intermediates, at different excitation
wavelengths. In the case of Mn, the higher-energy band was
found to associate with formation of singlet η5-CpMn(CO)2,
while the lower-energy band was found to associate with that
of triplet η5-CpMn(CO)2. Interactions of solvent molecules with
the singlet and triplet Mn complexes were also discussed.
The metal dicarbonyls η5-CpMn(CO)2 and η5-CpRe(CO)2
resulting from UV photolysis of a CO ligand were found to be
solvated by a solvent molecule in a few picoseconds. From
this point on, both metal complexes share a common reaction
pattern, in which solvation through chemically different sites
of a solvent molecule further partitions the reaction into two
routes. The final product could form on the ultrafast regime
when the coordinatively unsaturated 16-e- metal complex
interacted with the Si-H bond of the solvent molecule. On
longer time scales, both reactions (Mn and Re) were shown to
proceed through a kinetically favored intermediatesa metal
dicarbonyl solvated by the ethyl moiety of solvent molecule.
The branching ratio for these solvation-partitioned reaction
pathways, derived from the product kinetic traces taken at nanoor microsecond time scales, was argued to be governed by steric
interaction. The apparent rate-determining step was found to
originate from dissociative rearrangement from this ethyl solvate
to the thermodynamically more stable hydridosilyl product. The
associated free-energy barriers were also determined and were
in excellent agreement with results from previous low-temperature studies (for Mn).
Through detailed mechanistic study, we were able to experimentally determine the Si-H bond-breaking step. The time
scale for such a process in a liquid environment was, for the
first time, measured to be 4.4 ps. The result allows an upper
bound for the time scale of the Si-H bond-activation step,
confirming ab initio theoretical predictions.58 This is in sharp
(58) (a) Koga, N.; Morokuma, K. J. Am. Chem. Soc. 1993, 115, 6883.
(b) Musaev, D. G.; Morokuma, K. J. Am. Chem. Soc. 1995, 117, 799.
Reaction Mechanism of Si-H Bond ActiVation
contrast to the analogous C-H bond-cleaving process, which
takes approximately 230 ns.4 In this case, the 8.3 kcal/mol
energy barrier corresponding to cleaving an alkane C-H bond
has been attributed to poor spatial overlap and energy match of
the molecular orbitals between an alkane C-H bond and a
transition metal center; the hydrogen atom and an alkyl group
have to separate before forming chemical bonds with the metal
center.58 Extension of the current work, including a close
interplay between experiment and theory, to other transition
metals reacting with different chemical bonds is expected to
provide a better understanding about the chemistry of bondcleavage reactions.
Acknowledgment. This work was supported by a grant from
the National Science Foundation. We thank R. G. Bergman
for many helpful discussions; K. Morokuma for suggestions on
the ab initio calculations; C. B. Moore for the use of the static
FTIR spectrometer; and J. Yeston and P. J. Alaimo for assistance
with handling of the samples.
VI. Appendix
In the ultrafast experiment of η5-CpRe(CO)3/Et3SiH, several
events that occur concurrently on the ∼100-ps time scale have
complicated the transient spectra in the frequency region from
1990 to 2030 cm-1. These events include the recovery of the
parent η5-CpRe(12CO)3 bleach due to reequilibration of the
vibrational population, bleach recovery of η5-CpRe(12CO)2(13CO), and the formation of the product η5-CpRe(CO)2(H)(SiEt3). Summarized below are these processes and their
associated frequencies:
1990 cm-1:
product formation,
vibrational relaxation of 13V ) 2
2004 cm-1:
vibrational relaxation of 13V ) 1
2015 cm-1:
bleach recovery of 13V ) 0,
vibrational relaxation of 12V ) 1
2030 cm-1:
bleach recovery of 12V ) 0
where superscripts 12 and 13 refer respectively to the parent
molecule η5-CpRe(12CO)3 and the naturally abundant 13CO
J. Am. Chem. Soc., Vol. 120, No. 39, 1998 10165
monosubstituted η5-CpRe(13CO)(12CO)2. To deconvolve the
dynamics of individual events out of the complicated spectra,
we first consider the population of the ith vibrational state of
η5-CpRe(13CO)(12CO)2. Assuming that the relaxation rate of
the i state to the i - 1 state is proportional to the population ni,
and that ni is replenished only by relaxation from the i + 1
state, we can write the coupled rate equations for vibrational
states i ) 0, 1, and 2 as
d(13ni)
) -13ki,i-113ni + 13ki+1,i13ni+1
dt
(6)
where ki,j s are the rate constants for the transition from i state
to j state. Similarly, the population dynamics for CO vibrational
states i ) 0 and 1 of η5-CpRe(12CO)3 can be expressed as
d(12n0) 12 12
) k1,0 n1
dt
(7)
d(12n1)
) -12k1,012n1
dt
(8)
Finally, the formation of the final product at 1990 cm-1 is
assumed to be represented by a single exponential,
Iproduct ) Cproduct(1 - e-t/τproduct)
(9)
We further assume that the IR absorbance of state i, denoted
by Ii, is proportional to its population ni. The observed kinetic
traces can then be fit to this model by solving the coupled
differential equations (eqs 6-8) numerically and collecting terms
contributing to the relevant frequencies. The optimized parameters are as follows: 13I03 ) 0.29 ( 0.15, 13I02 ) 1.30 ( 0.015,
13I0 ) 0.92 ( 0.070, 13I0 ) -3.24 ( 0.16, 13k
3,2 ) 0.083 (
1
0
0.0036 ps-1, 13k2,1 ) 0.088 ( 0.0076 ps-1, 13k1,0 ) 0.088 (
0.015 ps-1, 12k1,0 ) 0.032 ( 0.0025 ps-1, and τproduct ) 4.35 (
2.6 ps. The deconvolved kinetic traces are displayed in Figure
6 and explained in the figure caption. Note that the intensities
at t ) 0 of state i, denoted by I0i, do not reflect the initial
population on each vibrational state, since we do not have a
priori knowledge of the extinction coefficient for each vibrational state.
JA980692F